Wingsuit Flight Motion Tracking

content

• 1.Import Libraries
• 2.Load Video & Extract Frames
  -- 2.1 Load Video
  -- 2.2 Extract key video properties
  -- 2.3 Extract and store frames
  -- 2.4 Release and display
• 3.Dynamic Motion Tracking with Optical Flow
  -- 3.1 Define feature detection parameters
  -- 3.2 Define LK optical flow parameters
  -- 3.3 Track features and draw color-doded motion vectors
• Generate & Display Animation
• Acknowledgment

Import libraries

import cv2
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.animation import FuncAnimation
from IPython.display import HTML
import os
import time
import matplotlib.animation as animation

Load Video & Extract Frames

Load video

video_cap=cv2.VideoCapture("wingsuit flight.mp4") 

Extract key video properties

#Extract file size
file_size=os.path.getsize("wingsuit flight.mp4") / (1024 * 1024)  # file size in MB
#Extract dimension
frame_width = int(video_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
frame_height = int(video_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#Extract total frames
total_frames = int(video_cap.get(cv2.CAP_PROP_FRAME_COUNT))
#Extract fps
frame_fps=video_cap.get(cv2.CAP_PROP_FPS)
#Extract duration
duration_sec = total_frames / frame_fps if frame_fps else 0
duration_min = duration_sec / 60
#Extract codec
codec = int(video_cap.get(cv2.CAP_PROP_FOURCC))
codec_str = "".join([chr((codec >> 8 * i) & 0xFF) for i in range(4)])
#Display video properties
print(f"File Size: {file_size:.2f} MB")
print(f"Dimensions: {frame_width}x{frame_height}")
print(f"Total Frames: {total_frames}")
print(f"FPS: {frame_fps}")
print(f"Duration: {duration_min:.2f} min({duration_sec:.2f})s ")
print(f"Codec: {codec_str}")

File Size: 1.43 MB
Dimensions: 640x360
Total Frames: 92
FPS: 9.0
Duration: 0.17 min(10.22)s 
Codec: h264

Extract and store frames

#Extract frames
frames_ori = []
frames_gray=[]
frames_max=90
for i in range(min(frames_max,total_frames)):
    re,frame=video_cap.read()
    if not re:
        break
    frames_ori.append(frame)
    frames_gray.append(cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY))

Release and display

#Release video capture object
video_cap.release()
#Count frames
print(f"Extracted {len(frames_ori)} frames.")
#Display first frame
frames_ori[0]=cv2.cvtColor(frames_ori[0],cv2.COLOR_BGR2RGB)
plt.imshow(frames_ori[0])
plt.axis('off') 
plt.show()

Extracted 90 frames.

Dynamic Motion Tracking with Optical Flow

Define feature detection parameters

feature_params = dict(
    maxCorners=200,        
    qualityLevel=0.01,     
    minDistance=10,        
    blockSize=15
)

Define LK optical flow parameters

lk_params = dict(
    winSize=(25, 25),
    maxLevel=5,
    criteria=(cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 30, 0.01)
    )

Track features and draw color-doded motion vectors

#Detect features in the first frame
p0 = cv2.goodFeaturesToTrack(frames_gray[0], mask=None, **feature_params)
p0 = p0.astype(np.float32)
pre=p0
#Define a function to remove duplicate points
def remove_duplicates(existing_points, new_points, min_distance=5):
    if existing_points is None or len(existing_points) == 0:
        return new_points
    if new_points is None or len(new_points) == 0:
        return np.empty((0, 1, 2), dtype=np.float32)
    
    existing_points = existing_points.reshape(-1, 2)
    new_points = new_points.reshape(-1, 2)
    
    filtered_new_points = []
    for new_pt in new_points:
        distances = np.linalg.norm(existing_points - new_pt, axis=1)
        if np.all(distances >= min_distance):
            filtered_new_points.append(new_pt)
    
    return np.array(filtered_new_points)

track_frames=[]
for i in range(1, len(frames_gray)):
    p_cur,status,err=cv2.calcOpticalFlowPyrLK(frames_gray[i-1],frames_gray[i],pre,None,**lk_params)
    good_new=p_cur[status==1]
    good_old=pre[status==1] 
    print(f"Frame {i}: Tracked points: {len(good_new)}")
    if len(good_new)<=50 or (i+1)%5==0:
        if(len(good_new)>50):
            p_new=cv2.goodFeaturesToTrack(frames_gray[i], mask=None, **feature_params)
            if p_new is not None:
                p_new=remove_duplicates(good_new,p_new)
                good_new=np.vstack((good_new,p_new))
                if len(good_new)>200:
                    good_new=good_new[:200]
        else:
            while(len(good_new)<=50):
                p_new=cv2.goodFeaturesToTrack(frames_gray[i], mask=None, **feature_params)
                if p_new is None:
                    break
                p_new=remove_duplicates(good_new,p_new)
                if len(p_new)==0:
                    break
                good_new=np.vstack((good_new,p_new))
                if len(good_new)>200:
                    good_new=good_new[:200]
        print(f"Frame {i}: Added new points, total now: {len(good_new)}")
    frame=frames_ori[i].copy()
    #Convert to color for visualization
    for new,old in zip(good_new,good_old):
        a,b = new.ravel()
        c,d = old.ravel()
        dx=a-c
        dy=b-d
        speed=np.sqrt(dx**2+dy**2)
        if speed<5:
            color=(255,0,0)  #Blue for slow
        elif speed<15:
            color=(0,255,0) #Green for medium
        else:
            color=(0,0,255)  #Red for fast
        frame=cv2.arrowedLine(frame, (int(c),int(d)), (int(a),int(b)), color, 2, tipLength=0.3)#Draw arrowed line
    track_frames.append(frame) 
    pre=good_new.reshape(-1,1,2)

Frame 1: Tracked points: 193
Frame 2: Tracked points: 185
Frame 3: Tracked points: 151
Frame 4: Tracked points: 150
Frame 4: Added new points, total now: 200
Frame 5: Tracked points: 199
Frame 6: Tracked points: 199
Frame 7: Tracked points: 196
Frame 8: Tracked points: 192
Frame 9: Tracked points: 192
Frame 9: Added new points, total now: 200
Frame 10: Tracked points: 200
Frame 11: Tracked points: 200
Frame 12: Tracked points: 199
Frame 13: Tracked points: 196
Frame 14: Tracked points: 192
Frame 14: Added new points, total now: 200
Frame 15: Tracked points: 196
Frame 16: Tracked points: 130
Frame 17: Tracked points: 122
Frame 18: Tracked points: 83
Frame 19: Tracked points: 80
Frame 19: Added new points, total now: 200
Frame 20: Tracked points: 197
Frame 21: Tracked points: 193
Frame 22: Tracked points: 170
Frame 23: Tracked points: 120
Frame 24: Tracked points: 103
Frame 24: Added new points, total now: 200
Frame 25: Tracked points: 184
Frame 26: Tracked points: 182
Frame 27: Tracked points: 167
Frame 28: Tracked points: 141
Frame 29: Tracked points: 137
Frame 29: Added new points, total now: 200
Frame 30: Tracked points: 169
Frame 31: Tracked points: 160
Frame 32: Tracked points: 158
Frame 33: Tracked points: 137
Frame 34: Tracked points: 135
Frame 34: Added new points, total now: 200
Frame 35: Tracked points: 189
Frame 36: Tracked points: 183
Frame 37: Tracked points: 175
Frame 38: Tracked points: 171
Frame 39: Tracked points: 163
Frame 39: Added new points, total now: 200
Frame 40: Tracked points: 186
Frame 41: Tracked points: 169
Frame 42: Tracked points: 168
Frame 43: Tracked points: 166
Frame 44: Tracked points: 164
Frame 44: Added new points, total now: 200
Frame 45: Tracked points: 196
Frame 46: Tracked points: 192
Frame 47: Tracked points: 190
Frame 48: Tracked points: 180
Frame 49: Tracked points: 167
Frame 49: Added new points, total now: 200
Frame 50: Tracked points: 194
Frame 51: Tracked points: 187
Frame 52: Tracked points: 170
Frame 53: Tracked points: 166
Frame 54: Tracked points: 165
Frame 54: Added new points, total now: 200
Frame 55: Tracked points: 199
Frame 56: Tracked points: 168
Frame 57: Tracked points: 165
Frame 58: Tracked points: 161
Frame 59: Tracked points: 158
Frame 59: Added new points, total now: 200
Frame 60: Tracked points: 198
Frame 61: Tracked points: 188
Frame 62: Tracked points: 164
Frame 63: Tracked points: 148
Frame 64: Tracked points: 138
Frame 64: Added new points, total now: 200
Frame 65: Tracked points: 191
Frame 66: Tracked points: 182
Frame 67: Tracked points: 179
Frame 68: Tracked points: 166
Frame 69: Tracked points: 161
Frame 69: Added new points, total now: 200
Frame 70: Tracked points: 185
Frame 71: Tracked points: 180
Frame 72: Tracked points: 169
Frame 73: Tracked points: 166
Frame 74: Tracked points: 159
Frame 74: Added new points, total now: 200
Frame 75: Tracked points: 196
Frame 76: Tracked points: 184
Frame 77: Tracked points: 172
Frame 78: Tracked points: 166
Frame 79: Tracked points: 161
Frame 79: Added new points, total now: 200
Frame 80: Tracked points: 195
Frame 81: Tracked points: 188
Frame 82: Tracked points: 172
Frame 83: Tracked points: 169
Frame 84: Tracked points: 166
Frame 84: Added new points, total now: 200
Frame 85: Tracked points: 197
Frame 86: Tracked points: 187
Frame 87: Tracked points: 184
Frame 88: Tracked points: 169
Frame 89: Tracked points: 160
Frame 89: Added new points, total now: 200

Generate & Display Animation

def update(frame):   #Define update function for animation
    plt.cla()
    frame_img=cv2.cvtColor(track_frames[frame],cv2.COLOR_BGR2RGB)
    img=plt.imshow(frame_img)
    plt.axis('off' )
    plt.title(f'Frame {frame+1}\n(Red = Fast Speed, Green=Medium Speed,Blue = Slow Speed)',fontsize=16)
    return []
#Create animation
plt.close('all')
fig=plt.figure(figsize=(10,6))
ani = animation.FuncAnimation(
    fig, update,
    frames=range(len(track_frames)),
    interval=1000/frame_fps,  
    blit=True,
    repeat=True
)
display(HTML(ani.to_html5_video() ))
ani.save('Wingsuit_Analysis.mp4', fps=frame_fps,codec='libx264')# Save animation as MP4
plt.close(fig)

Acknowledgement

The video footage is sourced from the YouTube video 'GoPro: Graham Dickinson's Insane Wingsuit Flight - Front Helmet Cam 2 of 3' (URL:https://www.youtube.com/watch?v=kTzwvyzzfjk). This material is used for educational purposes only, with no copyright infringement intended.